readtext.m

读入一个文本数据

	% Keep quoted text together.
	if(~isempty(opts.quotes))									% Should we look for quotes?
		if((length(opts.quotes) == 1) || (opts.quotes(1) == opts.quotes(2)))	% Opening char == ending.
			exclE=			find(text == opts.quotes(1));
			exclS=			exclE(1:2:end);
			exclE=			exclE(2:2:end);
		else													% Opening char ~= closing.
			exclS=			find(text == opts.quotes(1));
			exclE=			find(text == opts.quotes(2));
		end
		if((length(exclS) ~= length(exclE)) || any(exclS > exclE))
			close(th);											% Close waitbar or it'll linger.
			error('Opening and closing quotes don''t match in %s.', opts.fname); 
		elseif(~isempty(exclS))									% We do have quoted text.
			mywaitbar(0, th, '(readtext) Doing quotes...');		% Inform user.
			r=		1;
			rEnd=	length(exclS);
			n=		1;
			nEnd=	length(delimS);
			result.quote=	rEnd;
			exclS(end+1)=	0;	% This and next lines are needed in cases with only one quote in text.
			exclE(end+1)=	0;
			while((n <= nEnd) && (r <= rEnd)) % "Remove" delimiters and newlines within quotes.
				while((r <= rEnd) && (delimS(n) > exclE(r))), r= r+1;	end	% Next end-quote after newline.
				while((n <= nEnd) && (delimS(n) < exclS(r))), n= n+1;	end	% Next newline after strart-quote.
				while((n <= nEnd) && (delimS(n) >= exclS(r)) && (delimS(n) <= exclE(r)))
					delimS(n)=	0;											% Newlines inside quote. 
					n=			n+1;
				end
				mywaitbar(n/nEnd);								% Update waitbar.
			end
			mywaitbar(1);
			delimE=	delimE(delimS > 0);
			delimS=	delimS(delimS > 0);
		end
	end
	delimS=		delimS-1;										% Last char before delimiter.
	delimE=		[1 delimE(1:end-1)+1];							% First char after delimiter.
	
	% Do the stuff: convert text to cell (and maybe numeric) array.
	mywaitbar(0, th, sprintf('(readtext) Processing ''%s''...', opts.fname));
	r=				1;
	c=				1;											% Presize data to optimise speed.
	data=			cell(length(divRow), ceil(length(delimS)/(length(divRow)-1)));
	nums=			zeros(size(data));							% Presize nums to optimise speed.
	nEnd=			length(delimS);								% Prepare for a waitbar.
	istextual=		strcmp(opts.format, 'textual');
	for n=1:nEnd
		temp= 			strtrim(text(delimE(n):delimS(n)));		% Textual item.
		data{r, c}= 	temp;									% Textual item.
		if(~istextual)
			lenT=			length(temp);
			if(lenT > 0)
				% Try to get 123, 123i, 123i + 45, or 123i - 45
				[a, count, errmsg, nextindex]=	sscanf(temp, '%f %1[ij] %1[+-] %f', 4);
				if(isempty(errmsg) && (nextindex > lenT))
					if    (count == 1),			nums(r, c)=		a;
					elseif(count == 2),			nums(r, c)=		a(1)*i;
					elseif(count == 4),			nums(r, c)=		a(1)*i + (44 - a(3))*a(4);
					else						nums(r, c)=		NaN;
					end
				elseif(regexpi(temp, '[^0-9eij \.\+\-]', 'once'))	% Remove non-numbers.
					nums(r, c)=		NaN;
				else
					nums(r, c)= 	s2n(temp, lenT);			% Some other kind of complex number.
				end
			else			nums(r, c)=	NaN;
			end
		end
		if(text(delimS(n)+1) == eol)							% Next row.
			result.min=		min(result.min, c);					% Find shortest row.
			result.max=		max(result.max, c);					% Find longest row.
			r=				r+1;
			c=				0;
			if(bitand(r, 15) == 0), mywaitbar(n/nEnd);	end		% Update waitbar.
		end
		c=				c+1;
	end
	
	% Clean up the conversion and do the result statistics.
	mywaitbar(0, th, '(readtext) Cleaning up...');				% Inform user.
	data=				data(1:(r-1), 1:result.max);			% In case we started off to big.
	if(~strcmp(opts.format, 'textual')), nums= nums(1:(r-1), 1:result.max);	end		% In case we started off to big.
	if(all(cellfun('isempty', data)))
		data= 			{};
		r=				1;
		result.min=		0;
		result.max=		0;
	else
		while((size(data, 2) > 1) && all(cellfun('isempty', data(end, :))))	% Remove empty last lines. 
			data=	data(1:end-1, :); 
			nums=	nums(1:end-1, :); 
			r=		r-1;
		end 
		while((size(data, 1) > 1) && all(cellfun('isempty', data(:, end))))	% Remove empty last columns. 
			data=	data(:, 1:end-1); 
			nums=	nums(:, 1:end-1); 
			c=		c-1;
		end 
	end
	result.rows=		r-1;
	empties=			cellfun('isempty', data);				% Find empty items.
	result.emptyMask=	empties;
	if(strcmp(opts.format, 'textual'))
		result.numberMask=	repmat(false, size(data));			% No numbers, all strings.
		result.stringMask=	~empties;							% No numbers, all strings.
		data(empties)=		{opts.op_empty};					% Set correct empty value.
	else
		if(isempty(data)),	result.numberMask=	[];
		else				result.numberMask=	~(isnan(nums) & ~strcmp(data, 'NaN'));	% What converted well.
		end
		if(strcmp(opts.format, 'numeric'))
			nums(empties)=		opts.op_empty;					% Set correct empty value.
			data=				nums;							% Return the numeric array.
			result.stringMask=	~(empties | result.numberMask);	% Didn't convert well: so strs.
		else
			data(result.numberMask)= num2cell(nums(result.numberMask));	% Copy back numerics.
			data(empties)=		{opts.op_empty};						% Set correct empty value.
			result.stringMask=	cellfun('isclass', data, 'char');	% Well, the strings.
		end
	end
	result.empty=		sum(result.emptyMask(:));				% Count empties.
	result.numberMask=	result.numberMask & ~result.emptyMask;	% Empties don't count.
	result.number=		sum(result.numberMask(:));				% Count numbers.
	result.stringMask=	result.stringMask & ~result.emptyMask;	% Empties don't count.
	result.string=		sum(result.stringMask(:));				% Count strings.
	
 	close(th);													% Removing the waitbar. 
end


function x= s2n(s, lenS)
	x=		NaN;

	% Try to get 123 + 23i or 123 - 23i
	[a,count,errmsg,nextindex] = sscanf(s,'%f %1[+-] %f %1[ij]',4);
	if(isempty(errmsg) && (nextindex > lenS))
		if(count == 4),				x=		a(1) + (44 - a(2))*a(3)*i;
		end
		return
	end

	% Try to get i, i + 45, or i - 45
	[a,count,errmsg,nextindex] = sscanf(s,'%1[ij] %1[+-] %f',3);
	if(isempty(errmsg) && (nextindex > lenS))
		if(count == 1),				x=		i;
		elseif(count == 3),			x=		i + (44 - a(2))*a(3);
		end
		return
	end

	% Try to get 123 + i or 123 - i
	[a,count,errmsg,nextindex] = sscanf(s,'%f %1[+-] %1[ij]',3);
	if(isempty(errmsg) && (nextindex > lenS))
		if(count == 1),				x=		a(1);
		elseif(count == 3),			x=		a(1) + (44 - a(2))*i;
		end
		return
	end

	% Try to get -i, -i + 45, or -i - 45
	[a,count,errmsg,nextindex] = sscanf(s,'%1[+-] %1[ij] %1[+-] %f',4);
	if(isempty(errmsg) && (nextindex > lenS))
		if(count == 2),				x=		(44 - a(1))*i;
		elseif(count == 4),			x=		(44 - a(1))*i + (44 - a(3))*a(4);
		end
		return
	end

	% Try to get 123 + 23*i or 123 - 23*i
	[a,count,errmsg,nextindex] = sscanf(s,'%f %1[+-] %f %1[*] %1[ij]',5);
	if(isempty(errmsg) && (nextindex > lenS))
		if(count == 5),				x=		a(1) + (44 - a(2))*a(3)*i;
		end
		return
	end

	% Try to get 123*i, 123*i + 45, or 123*i - 45
	[a,count,errmsg,nextindex] = sscanf(s,'%f %1[*] %1[ij] %1[+-] %f',5);
	if(isempty(errmsg) && (nextindex > lenS))
		if(count == 1),				x=		a;
		elseif(count == 3),			x=		a(1)*i;
		elseif(count == 5),			x=		a(1)*i + (44 - a(4))*a(5);
		end
		return
	end

	% Try to get i*123 + 45 or i*123 - 45
	[a,count,errmsg,nextindex] = sscanf(s,'%1[ij] %1[*] %f %1[+-] %f',5);
	if(isempty(errmsg) && (nextindex > lenS))
		if(count == 1),				x=		i;
		elseif(count == 3),			x=		i*a(3);
		elseif(count == 5),			x=		i*a(3) + (44 - a(4))*a(5);
		end
		return
	end

	% Try to get -i*123 + 45 or -i*123 - 45
	[a,count,errmsg,nextindex] = sscanf(s,'%1[+-] %1[ij] %1[*] %f %1[+-] %f',6);
	if(isempty(errmsg) && (nextindex > lenS))
		if(count == 2),				x=		(44 - a(1))*i;
		elseif(count == 4),			x=		(44 - a(1))*i*a(4);
		elseif(count == 6),			x=		(44 - a(1))*i*a(4) + (44 - a(5))*a(6);
		end
		return
	end

	% Try to get 123 + i*45 or 123 - i*45
	[a,count,errmsg,nextindex] = sscanf(s,'%f %1[+-] %1[ij] %1[*] %f',5);
	if(isempty(errmsg) && (nextindex > lenS))
		if(count == 5),				x=		a(1) + (44 - a(2))*i*a(5);
		end
		return
	end

	% None of the above cases.
	x=		NaN;

end


function emptywaitbar(varargin)
end